Microsoft's unified open-source framework for building AI agents and multi-agent systems, combining AutoGen's multi-agent patterns with Semantic Kernel's enterprise features into a single Python and .NET SDK.
Microsoft's enterprise platform for building AI agents integrated with Azure and Microsoft 365 — deploy agents across your Microsoft ecosystem.
Microsoft Agent Framework is what you get when Microsoft merges its two competing AI agent projects — AutoGen and Semantic Kernel — into one unified framework. AutoGen brought multi-agent conversation patterns (agents talking to each other to solve problems). Semantic Kernel brought enterprise plumbing (state management, telemetry, model abstraction, type safety). The combined framework does both.
It entered public preview in October 2025, with AutoGen and Semantic Kernel moving to maintenance mode. If you were building on either of those, this is where Microsoft wants you to go. The GA target is Q1 2026.
Free and open-source under MIT license. You pay nothing for the framework itself. Your costs come from the AI models you connect to it (Azure OpenAI, OpenAI API, local models) and any Azure services you use for hosting.
If you deploy agents through Azure AI Foundry Agent Service (which uses this framework under the hood), Azure's standard compute and model pricing applies.
Developers building AI agents in Python or C#/.NET who want a production-ready framework backed by Microsoft. It's particularly relevant for teams already in the Azure ecosystem, using Azure OpenAI, or building multi-agent systems where agents need to collaborate, hand off tasks, and maintain state across conversations.
This is the only major agent framework with first-class support for both Python and .NET. LangChain, CrewAI, and most competitors are Python-only. If your engineering team works in C#, this is effectively your only serious option for a structured agent framework. The merger of AutoGen and Semantic Kernel also means you get both dynamic agent orchestration (AutoGen's strength) and deterministic workflow control (Semantic Kernel's strength) in one API.
Microsoft Agent Framework solves a real problem: the confusion of choosing between AutoGen and Semantic Kernel. The unified API is cleaner, the documentation is improving, and the framework inherits battle-tested components from both predecessors. The risk is Microsoft's track record of framework churn — developers who invested in AutoGen or Semantic Kernel are now migrating again, and there's understandable frustration about that pattern. For new projects, this is the right starting point for .NET teams and a strong option for Python teams. For existing AutoGen/Semantic Kernel projects, migration is straightforward but still work. Wait for GA (Q1 2026) if stability matters more than features.
Was this helpful?
Microsoft Agent Framework is the right answer to the wrong question Microsoft created by having two competing agent frameworks. By merging AutoGen and Semantic Kernel, they've produced a genuinely strong unified framework — particularly for .NET developers who had no real alternative. The dual orchestration model (dynamic agents + deterministic workflows) is architecturally sound, and the checkpointing and protocol support show serious engineering investment. The concerns are legitimate: preview status, Microsoft's framework churn history, and a smaller ecosystem than LangChain. For .NET teams, this is the obvious choice. For Python teams, it competes seriously with LangChain. For everyone, wait for GA if stability matters.
Choose between agent orchestration (AutoGen-style LLM-driven agents that reason and collaborate dynamically) and workflow orchestration (Semantic Kernel-style deterministic pipelines with business logic). Use them separately or combine them — an agent can trigger a deterministic workflow, or a workflow can delegate a step to an autonomous agent.
Use Case:
A customer support system where a deterministic workflow handles ticket routing and SLA tracking, but delegates the actual response generation to an LLM-driven agent that reasons about the customer's issue.
Consistent APIs across both Python and C#/.NET — same concepts, same patterns, just different language idioms. This isn't a Python framework with a .NET wrapper; both are first-class citizens with dedicated SDKs.
Use Case:
An enterprise with a .NET backend that wants to add AI agents without introducing Python into their deployment pipeline. The .NET SDK lets them build agents that integrate natively with their existing services.
Built-in support for group chats (multiple agents discussing a problem), reflection (agents reviewing their own output), sequential handoffs, and parallel execution. Inherited from AutoGen's research-proven patterns.
Use Case:
A code review system where a 'reviewer' agent identifies issues, a 'fixer' agent proposes solutions, and a 'validator' agent checks the fixes — all coordinated through the framework's group chat pattern.
Save agent state at any point and restore it later. Useful for debugging long-running agent workflows, implementing retry logic, and creating reproducible test scenarios. 'Time-travel' lets you replay from any checkpoint.
Use Case:
Debugging why a multi-agent financial analysis went wrong by replaying from the checkpoint just before the error, with different model parameters or tool configurations.
Model Context Protocol (MCP) integration for connecting agents to external tools, and Agent-to-Agent (A2A) protocol support for cross-framework agent communication. This means your agents can use tools from the broader MCP ecosystem and communicate with agents built on other frameworks.
Use Case:
An agent built with Microsoft Agent Framework calling tools exposed by a LangChain-based service through MCP, or communicating with a Google ADK agent via A2A protocol.
Free
one-time
Custom
Ready to get started with Microsoft Agent Framework?
View Pricing Options →Microsoft Agent Framework works with these platforms and services:
We believe in transparent reviews. Here's what Microsoft Agent Framework doesn't handle well:
Weekly insights on the latest AI tools, features, and trends delivered to your inbox.
AI Agent Builders
Open-source Python framework that orchestrates autonomous AI agents collaborating as teams to accomplish complex workflows. Define agents with specific roles and goals, then organize them into crews that execute sequential or parallel tasks. Agents delegate work, share context, and complete multi-step processes like market research, content creation, and data analysis. Supports 100+ LLM providers through LiteLLM integration and includes memory systems for agent learning. Features 48K+ GitHub stars with active community.
AI Agent Builders
Google's open-source framework for building, evaluating, and deploying multi-agent AI systems with Gemini and other LLMs.
AI Agent Builders
SDK for building AI agents with planners, memory, and connectors. - Enhanced AI-powered platform providing advanced capabilities for modern development and business workflows. Features comprehensive tooling, integrations, and scalable architecture designed for professional teams and enterprise environments.
Multi-Agent Builders
Microsoft's open-source framework for building multi-agent AI systems with asynchronous, event-driven architecture.
No reviews yet. Be the first to share your experience!
Get started with Microsoft Agent Framework and see if it's the right fit for your needs.
Get Started →Take our 60-second quiz to get personalized tool recommendations
Find Your Perfect AI Stack →Explore 20 ready-to-deploy AI agent templates for sales, support, dev, research, and operations.
Browse Agent Templates →